You are on page 1of 22

Performance evaluation of intensity modulated optical OFDM system with digital baseband distortion

Evgeny Vanin
Acreo Netlab, Electrum 236, SE-16440 Kista, Sweden email: evgeny.vanin@acreo.se

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 1

Acknowledgments Acknowledgments
ALPHA team and the project coordinator Dr. Misha Popov

Dr. Jianming Tang

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 2

Outline Outline
Part 1 : Introduction Part 1 : Introduction
History History Basic principles of OFDM (DMT): Basic principles of OFDM (DMT): QAM mapping QAM mapping digital (DFFT generated) subcarriers, digital (DFFT generated) subcarriers, cyclic prefix (CP) cyclic prefix (CP) digital to analog / analog to digital converters (DAC/ADC) channel estimation channel estimation power and bit loading power and bit loading adaptive signal modulation adaptive signal modulation

Overview of potential applications of intensity-modulated (IM) optical OFDM in intensityaccess and home networks access and home networks transmission media laser sources laser sources

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 3

Part 2: Performance evaluation of IM optical OFDM system Part 2: Performance evaluation of IM optical OFDM system
Simplified schematic of the system Simplified schematic of the system Digital baseband distortion: signal clipping and quantization in DAC Digital baseband distortion: signal clipping and quantization in DAC Signal biasing for optical intensity modulation Additive receiver noise Additive receiver noise

Importance of signal clipping (why clipping distortion is deliberate?) Importance of signal clipping (why clipping distortion is deliberate?)

Analytical results: Definition of system model parameters Definition of system model parameters Definition of system model parameters Definition of system model parameters Effective SNR and system BER versus clipping ratio and bit resolution Effective SNR and system BER versus clipping ratio and bit resolution Analytical model verification: comparison with direct error counting simulations Analytical model verification: comparison with direct error counting simulations Optimum clipping ratio Optimum clipping ratio Minimum required bit resolution versus QAM constellation size Optical power sensitivity Optical power sensitivity Comparison of IM optical OFDM and M-ary ASK systems Comparison of IM optical OFDM and M-ary ASK systems Conclusions Conclusions

(two alternative modulation formats for transmitting many bits per signal sample / bandwidth)

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 4

History of OFDM History of OFDM

R. W. Chang, Synthesis of band-limited orthogonal signals for multichannel data transmission, Bell Sys. Tech. J. 45, 1775-1796, 1966.

R. W. Chang, Orthogonal frequency multiplex data transmission system, U.S. Patent 3 488 445, 1970.

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 5

References.: J. Armstrong, OFDM for Optical Communications, J. Lightwave Technol. 27, 189-204, (2009). W. Shieh, I. Djordjevic, OFDM for optical communications (Academic Press, 2010). Abbreviations: ADSL - Asymmetric Digital Subscriber Line, DMT - Discrete MultiTone

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 6

Why OFDM? Why OFDM?

In RF communication systems OFDM provides a simple way of dealing with multipath transmission by using a relatively simple DSP algorithms

In optical communications OFDM is useful for: compensation of modal dispersion in MMF, ... chromatic dispersion in SMF, ... multipath in free space effective use of bandwidth limited opto-electronic components ...

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 7

Basic principles of OFDM Basic principles of OFDM


OFDM (DMT) transmitter
(real valued signal, single DAC, intensity modulation, double sideband) QAM mapping/coding S/P CP

I IFFT

Add Cyclic prefix

P/S

OFDM symbol

IFFT

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 8

DC bias

OFDM symbol sequence DAC


Laser source

Digital Analog

Optical signal Transmission (SMF, MMF, POF, free space)

Tx OFDM spectrum

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 9

OFDM receiver
Rx OFDM spectrum

CP

Optical signal

Synchronized OFDM symbol sequence PD ADC

S/P

Remove Cyclic prefix

FFT

Equalizer

P/S Demaping/ ecoding Data extraction

Channel estimation Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 10

Power and bit loading. Adaptive modulation.


What if signal spectrum distorted by frequency selective fading?
Tx
1x10 Spectrum (a.u.) 1x10 1x10 1x10 1x10 -8 -4 0 4 frequency, GHz 8
-4 -3 -2 -1 0

1x10 Spectrum (a.u.) 1x10 1x10 1x10 1x10


-4 -3 -2 -1

Rx

-8

-4 0 4 frequency, GHz

Solution: Solution: - Power loading: redistribute the signal power among subcarriers at DAC input - Bit loading: use different size of QAM constellation among subcarriers - Adaptive modulation: adjust power and bit size at each subcarrier to maximize the transmission capacity
1x10 Spectrum (a.u.) 1x10 1x10 1x10 1x10
-2 -1 0

1x100

x
Spectrum (a.u.) 1x10
-1

Rx
1x10
-2

-3

1x10
-4

-3

1x10 -8 -4 0 4 frequency, GHz 8

-4

-8

-4 0 4 frequency, GHz

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 11

Potential applications of IM optical OFDM in access and home networks


Access networks (100m 20km) Transmission media: Transmission media: single-mode fibre single multi-mode silica glass fibre multi-mode silica glass fibre Laser sources: Laser sources: directly modulated laser diodes directly modulated laser diodes reflective semiconductor amplifiers reflective semiconductor amplifiers vertical cavity surface-emitting lasers surfaceHome networks (up to 50m) Transmission media: plastic optical fibre free space (Visible Light Communications) Laser sources: light emitting diodes

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 12

Part 2: Performance evaluation of IM optical OFDM system


Simplified schematic of the system Simplified schematic of the system Digital baseband distortion: signal clipping and quantization in DAC Digital baseband distortion: signal clipping and quantization in DAC Signal biasing for optical intensity modulation Signal biasing for optical intensity modulation Additive receiver noise

Importance of signal clipping (why clipping distortion is deliberate?) Importance of signal clipping (why clipping distortion is deliberate?)

Analytical results: Analytical results: Definition of system model parameters Effective SNR and system BER versus clipping ratio and bit resolution Effective SNR and system BER versus clipping ratio and bit resolution Optimum clipping ratio Optimum clipping ratio Minimum required bit resolution versus QAM constellation size Minimum required bit resolution versus QAM constellation size Optical power sensitivity of OFDM receiver Optical power sensitivity of OFDM receiver Comparison of IM optical OFDM and M-ary ASK systems Comparison of IM optical OFDM and M-ary ASK systems Conclusions

(two alternative modulation formats for transmitting many bits per signal sample / bandwidth) (two alternative modulation formats for transmitting many bits per signal sample / bandwidth)

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 13

Simplified schematic of IM optical OFDM system


DC bias Optical fiber Laser source PD ADC
DATA

DATA

DSP DAC

DSP

Inset A:
Digital baseband distortion due to clipping and quantization

Inset B:

Laser source modulation

Optical power

DC bias

Drive current

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 14

Importance of signal clipping


Optical power

Average optical power level

DC bias
time

Useful signal power is small.

Electrical OFDM signal

time

DC bias + Clipping Clipping


time

Useful signal power is increased. Clipping distortion Clipping distortion noise is small.

OFDM disadvantage: high peak-to-average power ratio (PAPR)


- PAPR is proportional to the number of subcarriers - Probability of high-power peaks is low

DC bias + Extra Clipping


time

Clipping distortion noise is too large

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 15

Definition of system model parameters


x (t )
Clipping ratio:

R=
x0

x0 < x 2 (t ) >

x0
time

ADC bit resolution:

N bit

Number of quantization levels:


2 I signal C &Q

2 Nbit
I2 I noise C &Q + I therm + I shot + I2 + I2

Effective signal to noise ratio:

SNReff =
2

Modified total SNR:

x ( P0 ) S0 = 0 2 xDC (Sth + 2q P0 ) f
2

Received average optical power:

< Popt >= P0

time

Analytical derivations are based on Bussgang theorem: J. J. Bussgang, Crosscorrelation functions of amplitude-distorted Gaussian signals, Research Lab. Electron, M.I.T., Cambridge, MA, USA, Tech. Rep. 216, Mar. 1952.

For detail derivations see: E. Vanin, Optics Express, Vol. 19, Issue 5, pp. 4280-4293 (2011)

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 16

Effective SNR versus clipping ratio


50 40 SNReff, dB 30 20 10 0 0 2 4 6 8 Clipping ratio, dB 10 12 14
20dB 15dB 10dB 30dB 25dB

SNReff @ optimum clipping ratio


55dB 50dB 45dB 40dB 35dB

S0

BER for rectangular M-QAM:

BERM QAM

3 SNReff 2 1 = 1 erfc log2 M M 2 ( M 1)


Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 17

Effective SNR versus clipping ratio and ADC bit resolution


50 SNReff @ optimum clipping ratio
No quantization

40 SNReff, dB 30 20 10 0 0 2 4 6 8 Clipping ratio, dB 10 12

S0=45dB

Nbit=8 Nbit=6 Nbit=5 Nbit=4 Nbit=3 Nbit=2 14


50

SNReff @ optimum clipping ratio


40

55dB

S0

50dB

45dB SNReff, dB
30

40dB

35dB 20

30dB

25dB 10

20dB

15dB

10dB 0 0 2 4 6 8

10

12

14

Clipping ratio, dB

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 18

BER versus clipping ratio


1x10-2

1x10-2

BER

1x10-3
16-QAM S0=28dB, Nbit=4 NFFT=16 NFFT=32 NFFT=128

BER 1x10-3

1x10-4 4 6 8 10 Clipping ratio, dB

1x10-4
12

64-QAM S 0=36dB, Nbit=5 NFFT=16 NFFT=32 NFFT=128

8 10 Clipping ratio, dB

12

1x10-2

BER

1x10-3

1x10-4 8

256-QAM S 0=44dB, Nbit=6 NFFT=16 NFFT=32 NFFT=128

10 12 Clipping ratio, dB
Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

14

Slide 19

System performance evaluation at optimum clipping ratio


50 40 S0, dB S0, dB 20 10 0 10 SNR , dB SNReff, dB Optimum clipping ratio, dB 1x10 1x10 BER 1x10-4 1x10-5
64QAM 256QAM 128QAM 32QAM 16QAM QPSK

Nbit=2 Nbit=6

Nbit=3

Nbit=4

Nbit=5

Nbit=7 Nbit=8

50 40 30

30 20 10 0 20 30 12 10 8 6 4 2 10 20 SNR eff, dB 30

-20

-16

-12

-8

-4

Optical power, dBm Optical power, dBm

-2

Optical receiver parameters

-3

Responsivity Thermal noise current Bandwidth

S
th

0.8 A/W

16 pA/ Hz

10 GHz

1x10 1x10
-7

-6

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 20

IM optical OFDM versus M-ary ASK


1x10-2
2.9dB 3.1dB 3.5dB

1x10

-3

3.3dB

1x10-4

BER

1x10-5

4-QAM/QPSK

16-QAM

256-QAM

1x10

-6

2-ASK/OOK

4-ASK

8-ASK 16-ASK

32-ASK 64-QAM

1x10-7 -20 -15 -10 -5

Optical power, dBm


Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 21

Conclusions

- Analytical results on performance evaluation of IM optical OFDM system - Analytical results on performance evaluation of IM optical OFDM system modelling are obtained modelling are obtained - Optimum clipping ratio - Optimum clipping ratio - BER - Received optical power sensitivity - Received optical power sensitivity - Minimum required bit resolution versus QAM constellation size - Minimum required bit resolution versus QAM constellation size

- IM optical OFDM requires about 3-3.5 dB higher received optical power in - IM optical OFDM requires about 3-3.5 dB higher received optical power in comparison with M-ary ASK system comparison with M-ary ASK system

- Considered model does not account for fibre dispersion and nonlinearity, laser chirp etc. Accounting for these effects is the prospect for future work. chirp etc. Accounting for these effects is the prospect for future work.

- Presented results can be used as an idealized reference for complicated - Presented results can be used as an idealized reference for complicated numerical codes and system experiments

Evgeny Vanin, Acreo Netlab Seminar, 2 March 2011

Slide 22

You might also like